Involuntary and voluntary patient motion is a major obstacle for achieving high-precision radiation therapy of cancers especially in the thorax and upper abdomen. As the target is continuously moving, an additional margin has to be added to the clinical target volume to compensate for the uncertainty caused by the respiration-induced organ motion, causing unnecessary toxicity to the normal tissue and limiting the dose delivered to the target. Due to the lack of reliable and safe direct tumor tracking methods, surrogates are commonly used in clinics to compensate for the tumor motion during radiotherapy. However, the relationship between the surrogate and tumor motion cannot be generalized as it depends on individual patients, tumor location, and treatment fractions. Therefore, our understanding of the correlation between the surrogate and the actual intrafraction tumor motion due to respiration is very limited. Although 4D-CT image is typically used for the radiotherapy motion assessment and planning, it is an oversimplified snapshot representation of a single-breathing cycle and is inherently limited by radiation dose. Consequently, these limitations prevent applicability in acquiring information over timescales that represent a treatment session. Advances in dynamic MRI present opportunities for the pre-treatment radiotherapy setting that have yet to be systematically harnessed. MRI is highly advantageous as it is non-ionizing and provides excellent soft tissue contrast. Variations in respiratory motion can be captured by imaging over longer durations or more frequently. MRI also achieves faster frame rates than 4D-CT, which likely provides a more accurate description of the moving anatomy. In this study, we will use dynamic MRI techniques to characterize tumor motion as well as surrogates used for predicting tumor motion. By way of this research, we will fully characterize patient-specific respiration- induced tumor motion and evaluate the accuracy and effectiveness of surrogate-based motion management strategies currently used in radiotherapy. Specifically, 4D-MRI obtained over a temporal duration consistent with radiotherapy treatments will provide spatio-temporal information of both the tumor and surrogate, and therefore can serve as a means to assess the quality of the tumor motion tracking with the surrogate, suggesting an appropriate motion management strategy for individual patients. The proposed methods will have general applicability to both thoracic and abdominal sites of disease. We intend that this research will be ground breaking and, if so, will likely result in a change of practice in radiotherapy, benefiting innumerable thoracic and abdominal cancer patients in the future.